Why Snapshot saves time - and money


What is the advantage of hyperspectral video cameras over regular hyperspectral imagers? What is the technical difference of this new kind of technology and how does it improve your application? We try to answer this and connected questions in the following article.


What is the advantage of hyperspectral snapshot imagers

The core advantage of hyperspectral snapshot imagers is the efficiency of usage of the incident light. Ordinary hyperspectral imagers have one thing in common. They use a small fraction of the incident light for the image generation.

On the one hand we have push-broom sensors, which restrict the light throughput of the objective of the camera by a micrometer-thin slit. On the other hand we see tunable filter devices, which only transmit a few nm of the full incident spectra onto the sensor. In either way, most of the incident light is blocked geometrically or is filtered out.

Snapshot imaging systems transmit the full spectrum of the light for the full image size. Thus, these kind of spectrometers capture a three dimensional dataset [x, y, lambda] in one sensor readout.

By this definition, Cubert GmbH is the only commercial source for hyperspectral snapshot cameras worldwide.

Why Snapshot - Snapshot Advantage - Cubert GmbH - Germany

Figure 1: Comparison of different imaging spectrometer technology

  • a) Tunable filter spectrometers capture one wavelength at a given time. The system has to switch through the different channels to capture one spectral cube, by combining multiple images. These images always have a temporal offset to each other.
  • b) Push-broom spectrometers capture one spectral slice per image readout. To generate a hyperspectral cube, the device has to be scanned over the object.
  • c) The hyperspectral snapshot imager captures x, y and lambda in the same sensor read out. The datacube is taken at once without any timely offset.

Advantages of the Hyperspectral Snapshot Technology

There are several advantages of the snapshot imagers. Most obvious is the easy image generation of the devices.

Since the whole three-dimensional dataset is captured with one sensor readout, there is no need for image combination of any kind to generate the hyperspectral cube.

Thus we do observe no moving artifacts and all channels as well as all locations are captured at the same instance of time. This also enables capturing very fast processes, making hyperspectral high-speed cameras with integration times of µs and less possible (link).

Due to the speed of the capturing, snapshot hyper spectrometers by definition can have no moving parts, making the whole setup sturdy and robust.

  • no moving parts
  • no moving artifacts
  • easy-to-use
  • real hyperspectral video camera
  • high light throughput

Since hyperspectral snapshot imagers can be build in a way, which transmits the full spectra of the complete image at the same time, these devices offer a significant speed improvement in integration time in comparison to ordinary hyperspectral imagers.

Hyperspectral snapshot imagers offer superior Integration time (in same light condition) and thus unbeaten Signal-to-Noise.

Speed Improvement

The speed improvement can be calculated like described in Review of snapshot spectral imaging technologies. By this, we can calculate the amount of voxels and thus the corresponding speed improvement.

As a Pixel describes an element of a 2D detector array, a Voxel describes a Voleum element of a 3D measurement device. In our case the size of the Voxel is governed by the spatial resolution in x, y, and the spectral resolution in wavelength. For a camera having 50×50 spatial pixels and 125 bands, the total number of Voxels is 50x50x125 = 312,500 per image capture.

In a typical snapshot setup, this 312,500 Voxels are measured in one capture event, whereas in an push-broom setup, only 50×125 = 6,250 Voxels can be captured in one frame. Thus the speed of the push-broom device is 50 times smaller than the snapshot imaging setup.

The  tunable filter setup on the other hand will be 125 times slower than the snapshot spectrometer, since an image has to be generated for each channel.

Enquire about a product or application

Spectral Skin Analysis
Hyperspectral Vegetation Indices